The Agent Skills Directory

[SAFE]: The skill primarily contains documentation and workflow patterns for multi-agent coordination. It does not contain executable scripts or malicious instructions.
[DATA_EXPOSURE]: The skill references local endpoints (localhost:3110 and localhost:18789) for task tracking and routing. These are internal to the orchestration environment and do not involve unauthorized external data transmission.
[INDIRECT_PROMPT_INJECTION]: The orchestration patterns involve passing data between agents (e.g., from CCC to Alfred). While this creates a potential surface for indirect injection where one agent's output could influence another, the skill itself is a high-level coordination guide and does not implement insecure processing of untrusted external data.

openclaw-coordination